```

feat(engine): 优化智能体循环中的助手消息处理逻辑

- 在没有工具调用时才添加助手消息到上下文
- 确保工具调用响应正确添加到消息上下文中
- 修复了消息构建的条件逻辑

fix(cron): 改进定时任务调度的时间解析功能

- 添加正则表达式导入用于时间显示解析
- 实现从显示文本中提取毫秒间隔的功能
- 增强整数转换的安全性，避免类型错误
- 优化定时任务配置的解析逻辑

feat(outlook): 增强Outlook集成的功能和稳定性

- 将默认超时时间从10秒增加到180秒
- 为状态检查函数添加可选的验证参数
- 串行执行邮件概览获取操作而非并行
- 改进连接状态验证逻辑

feat(channel): 添加设备名称作为会话标识的选项

- 为终端WebSocket适配器添加新的配置选项
- 实现基于设备名称生成会话对等ID的功能
- 记录原始对等ID和设备名称的元数据
- 支持从设备名称创建会话对等ID

feat(skills): 完善技能学习评估系统和进度跟踪

- 在应用启动时自动调度待评估的技能草稿
- 为技能评估工作创建独立的循环工厂
- 实现异步技能评估任务的取消和清理机制
- 添加技能评估进度报告和状态跟踪功能
- 扩展会话列表API以包含更多详细信息
- 防止对不存在的会话进行操作
- 优化技能草稿提交和评估的业务逻辑

perf(skills): 提升技能评估的并发性能

- 实现并行技能案例评估以提高效率
- 添加最大并行案例数的环境变量控制
- 实现实时评估进度更新和回调机制
- 优化评估过程中的资源管理和同步

refactor(services): 创建隔离的智能体循环实例

- 添加创建独立智能体循环的工厂方法
- 确保新循环继承运行时服务配置
- 支持技能评估等需要隔离环境的场景
```

app-instance/backend/tests/unit/test_skill_learning_eval.py

@@ -201,6 +201,22 @@ class FakeReplayRunner:
         }
 
 
+class ConcurrentReplayRunner(FakeReplayRunner):
+    def __init__(self) -> None:
+        super().__init__()
+        self.active = 0
+        self.max_active = 0
+
+    async def run_arm(self, request):
+        self.active += 1
+        self.max_active = max(self.max_active, self.active)
+        await asyncio.sleep(0.02)
+        try:
+            return await super().run_arm(request)
+        finally:
+            self.active -= 1
+
+
 def test_eval_report_includes_replay_case_and_coverage(tmp_path: Path) -> None:
     pipeline = _pipeline(tmp_path)
     draft = pipeline.draft_service.create_new_skill_draft(
@@ -238,6 +254,94 @@ def test_eval_report_includes_replay_case_and_coverage(tmp_path: Path) -> None:
     assert report.tool_execution_summary["score_role"] == "diagnostic_only"
 
 
+def test_replay_eval_reports_arm_progress(tmp_path: Path) -> None:
+    pipeline = _pipeline(tmp_path)
+    draft = pipeline.draft_service.create_new_skill_draft(
+        skill_name="release-checklist",
+        proposed_content="# Release\n\nRun tests.",
+        proposed_frontmatter={"description": "release", "tools": []},
+        created_by="test",
+        reason="test",
+    )
+    pipeline.learning_store.update_learning_candidate(
+        "candidate-1",
+        draft_skill_name=draft.skill_name,
+        draft_id=draft.draft_id,
+    )
+    progress: list[dict] = []
+
+    asyncio.run(
+        pipeline.evaluate_draft(
+            "candidate-1",
+            draft.skill_name,
+            draft.draft_id,
+            provider_bundle=_bundle(),
+            replay_runner=FakeReplayRunner(),
+            progress_callback=progress.append,
+        )
+    )
+
+    assert progress[0] == {
+        "phase": "replaying",
+        "completed_arms": 0,
+        "total_arms": 20,
+        "completed_cases": 0,
+        "total_cases": 10,
+    }
+    assert progress[-1] == {
+        "phase": "replaying",
+        "completed_arms": 20,
+        "total_arms": 20,
+        "completed_cases": 10,
+        "total_cases": 10,
+    }
+
+
+def test_replay_eval_runs_cases_with_bounded_parallelism(tmp_path: Path) -> None:
+    pipeline = _pipeline(tmp_path)
+    pipeline.evaluator = SkillDraftEvaluator(
+        pipeline.learning_service.run_store,
+        max_parallel_cases=2,
+    )
+    draft = pipeline.draft_service.create_new_skill_draft(
+        skill_name="release-checklist",
+        proposed_content="# Release\n\nRun tests.",
+        proposed_frontmatter={"description": "release", "tools": []},
+        created_by="test",
+        reason="test",
+    )
+    pipeline.learning_store.update_learning_candidate(
+        "candidate-1",
+        draft_skill_name=draft.skill_name,
+        draft_id=draft.draft_id,
+    )
+    replay_runner = ConcurrentReplayRunner()
+
+    report = asyncio.run(
+        pipeline.evaluate_draft(
+            "candidate-1",
+            draft.skill_name,
+            draft.draft_id,
+            provider_bundle=_bundle(),
+            replay_runner=replay_runner,
+        )
+    )
+
+    assert replay_runner.max_active == 2
+    assert [case["run_id"] for case in report.cases] == [
+        "run-1",
+        "synthetic:candidate-1:01",
+        "synthetic:candidate-1:02",
+        "synthetic:candidate-1:03",
+        "synthetic:candidate-1:04",
+        "synthetic:candidate-1:05",
+        "synthetic:candidate-1:06",
+        "synthetic:candidate-1:07",
+        "synthetic:candidate-1:08",
+        "synthetic:candidate-1:09",
+    ]
+
+
 def test_replay_main_score_uses_validator_not_tool_success(tmp_path: Path) -> None:
     pipeline = _pipeline(tmp_path)
     pipeline.learning_store.update_learning_candidate(